An anaphora is, roughly speaking, an abbreviated linguistic form whose full meaning can only be recovered by reference to the context; the reference is called Anaphora, and the mention of the entity to which anaphora refers is called the Antecedent.

It is difficult for machine (even human beings) to resolve anaphora - finding out it's antecedent. In general, there are three approaches for anaphora resolution. The first approach is called Syntactic Method, which uses syntactic information of anaphora (gender and number). because there is a strong constraint that anaphora and it's antecedent are coherent. This method is powerful, when an anaphora is a pronoun, especially a personal pronoun.

However, in the case of definite noun, the second approach called Semantic Method may be more convenient, because definite pronoun has more lexical information related to it's referent. Of course, if the perfect dictionary could be constructed. The meaning of a sentence depends on the context in which it is used with other sentences.

The third approach called Discourse Method is base on discourse factor such as focus and viewpoint. The mention of the entity which is focused tend to be referred in subsequent sentences. So the antecedent may be one of previously focused word. In most case, this could be going well. However, it is difficult to determine focused words in sentences. Another method based on discourse factor, uses the Viewpoint of sentences instead of the focus. Even if the verbs (predicates) of the two sentences are semantically linked, the connection between sentences will be felt unnatural if the relation between viewpoints is not established, and the sentences do not form a context. In viewpoint based method, Antecedents are determined so that the viewpoint relation should be established.